At first with reference to figure 3.In the prior art, the distribution control to information source, channel-encoded bit number is to liken Controlling Source to by the channel noise that channel SNR estimator feeds back to realize.Use the signal to noise ratio mode to represent the quality of signal at receiving terminal, and then the differentiation Channel Transmission condition of using, in general, be difficult to accurately differentiate and statement.Particularly be difficult to measure signal to noise ratio under the fading channel condition, the differentiation channel condition that at all can't use makes that control method is difficult to play a role.And the channel bit error rate that adopts receiving terminal to feed back to is made Controlling Source, because the measurement of the error rate is not subjected to the channel condition influence, more can accurately reflects the receiving end signal quality than the signal to noise ratio mode, thereby can differentiate channel condition exactly.
Principle of the present invention as described below, with reference to figure 1,2.Suppose that information source is input as the speech data or the pictorial data of sample quantization, data are to be frame input according to every T millisecond of time, and the bit number of each frame input is fixed, and is assumed to K
1The purpose of source encoding is the K to each incoming frame
1Bit data compresses, because compression is that the distortion compression is arranged, therefore the basic principle of compression is to make the data of compressed encoding can recover people's acceptable effect after channel decoding.Suppose K
1Coding after the data compression of bit becomes K
2Bit, K at this moment
2<K
1, owing to adopt different compression criterions, K
2Can be for several different values, as K
2=K
21, K
22, K
23..., K
2N, and, K
21<K
22<K
23<...<K
2N<K
1Simultaneously, chnnel coding is to K
2The data of bit are carried out chnnel coding, and the purpose of coding is that protected data is not subjected to channel effect, so must increase the information redundancy amount, the bit number behind the signaling channel coding is K
3, and K is arranged
3>K
2Because capability of communication system is certain, for making in the communication system that number of users is constant or based on other reason, the total bit number that is used for source encoding and chnnel coding is certain, i.e. K
3To message processing modules such as modulation is a constant.Thus, according to K
2Difference, be used for the bit number (K of chnnel coding
4=K
3-K
2) also become K thereupon
2Increase K
4Reduce.
During decoding, adopt channel bit error rate to detect and substitute the signal to noise ratio detection, can obtain the channel condition estimated value more exactly, and can embody the quality that the decoding end recovers information source better.By (can being the decoding of convolution code-VITERBI with channel decoding, TURBO sign indicating number-MAP or the decoding of SOVA class, or the interpretation method of block code etc.) preceding data and the data after the channel decoding carry out comparing behind the Gray code, obtain the instantaneous error rate estimated value of channel, after above-mentioned instantaneous error rate estimated value carried out smoothing processing, obtain the comprehensive estimated value of channel bit error rate, should comprehensive estimated value feed back to transmitting terminal again, distribution in order to the bit number of control source encoding and chnnel coding makes total decoding recover the better quality of information source data.
At this moment, the basic operation flow process of transmitting terminal is as follows:
1, information source data framing input, every frame K
1Bit;
2, K
1Bit data is by source encoding module boil down to K
2Bit, wherein,
K
2<K
1,
K
2=K
21,K
22,K
23,...,K
2N,
K
21<K
22<K
23<...<K
2N<K
1;
3, K
2The data of bit become K by channel coding module
3Bit is because K
2Change, so be used for the bit number K of chnnel coding
4=K
3-K
2Also change.
Because the general bit number K that is used for chnnel coding (refering in particular to convolution code, TURBO sign indicating number etc.)
4Be K
2Integral multiple, for example, 1 or 2 times; For making K
4=K
3-K
2, adopt multiplex technique commonly used in the communication system, as punching and repetition, regulate by the bit number behind the codings such as convolution code or TURBO sign indicating number, make its final K
4=K
3-K
2So the bit number of regulating chnnel coding here is actual in comprising multiplexing regulation technology.
4, send by channel after the coded modulation after will compressing;
Error code may take place through channel and be sent to receiving terminal in the data after the modulation.
The basic operation flow process of receiving terminal is as follows:
1, modulation back data are waited until the channel decoding module by demodulation;
2, carry out channel decoding and source decoding;
3, the data of the data after the channel decoding being carried out before Gray code and the channel decoding are compared, obtain the instantaneous error rate estimated value of channel, after above-mentioned instantaneous error rate estimated value carried out smoothing processing, obtain the comprehensive estimated value of channel bit error rate, the comprehensive estimated value of this channel bit error rate is fed back to the transmitting terminal control bit distribute.At first determine source encoding bit number K
2, method is: determine the source encoding bit number according to the scope of channel bit error rate parameter value, next utilizes system's information source, the total bit number of chnnel coding to deduct above-mentioned source encoding bit number, as the channel-encoded bit number.Determine source encoding bit number assigning process as:
As { the error rate<B
1K then
2=K
2N;
B
1<=the error rate<B
2K then
2=K
2N-1
B
N-1<=the error rate is K then
2=K
21}
Wherein: B
1<B
2...<B
N-1
Above-mentioned B
1, B
2..., B
N-1Value mainly be to determine according to speech decoding decoding performance under wrong code-bar spare, through the experiment decision.Basic principle is under current error code, regulates the bit number of chnnel coding and source encoding, selects best decoding voice quality.
The present invention can be applied to exist in the communication system of source encoding, chnnel coding.With reference to figure 4.Suppose that information source adopts the AMR speech coder and decoder, chnnel coding adopts convolutional encoding, VITERBI decoding is adopted in decoding, transmission channel adopts Rayleigh channel, receiving terminal adopts bit-error-rate estimator to detect the channel bit error rate parameter, transmitting terminal adopts the bit number of Bit Allocation in Discrete controller control source encoding and chnnel coding to distribute, phonetic entry is the voice PCM data of sample quantization, the 8KHz sample rate, 64Kbitps, the A-LAW compression, the AMR coding is speech data (160 points of a frame to every 20ms, the 160*16 bit) compresses, then K
1=2560 bits.
AMR has 8 speed class, is respectively: 4.75,5.15,5.90,6.70,7.40,7.95,10.2 or 12.2kbps, and corresponding various speed, every speech frame passes the speech bits number and is: 95,103,118,134,148,159,202,244 bits.Bit number behind the AMR coding is K
2, K
2=95,103,118,134,148,159,202,244.Simultaneously, to K
2Carry out convolutional encoding, the bit number behind the signaling channel coding is K
3=488 is constant.Thus, according to K
2Difference, be used for the bit number (K of chnnel coding
4=K
3-K
2) also become K thereupon
2Increase K
4Reduce.K
4=393,385,370,354,340,329,286,244。The speed of supposing AMR is 12.2kbps, and then to pass the speech parameter bit number be 244 bits to every speech frame.
During decoding, bit-error-rate estimator is estimated the preceding error rate of VITERBI decoding, method of estimation is estimated for data before the data after the VITERBI decoding are carried out Gray code and deciphered relatively obtain the instantaneous error rate, by the comprehensive error rate estimated value after instantaneous error rate estimated value is smoothly obtained handling.And this comprehensive error rate estimated value fed back to transmitting terminal Bit Allocation in Discrete controller, allow the bit number of Bit Allocation in Discrete controller control source encoding and chnnel coding distribute.
Corresponding transmitting terminal basic operation flow process is as follows:
1, information source data framing input, every frame 2560 bits;
2, K
1Bit data is by source encoding module boil down to K
2Bit, wherein,
K
2<2560,
K
2=95,103,118,134,148,159,202,244 bits;
3, K
2The data of bit become 488 bits by channel coding module, because K
2Bit number changes, so be used for the bit number K of chnnel coding
4=K
3-K
2Also change; If K
2=244 bits, then K
4=244 bits.
4, the data after will encoding send to channel through modulation.
Error code may take place through channel and be sent to receiving terminal in modulation back data.
Corresponding receiving terminal basic operation flow process is as follows:
1, modulation back data are waited until the channel decoding module by demodulation;
2, carry out channel decoding and source decoding;
3, the data of bit-error-rate estimator after with the channel decoding data of carrying out before Gray code and the channel decoding are compared, obtain the instantaneous error rate estimated value of channel, after above-mentioned instantaneous error rate estimated value carried out smoothing processing, obtain the comprehensive estimated value of channel bit error rate, the comprehensive estimated value of this channel bit error rate is fed back to transmitting terminal Bit Allocation in Discrete controller, source encoding bit number K is at first determined in the distribution of the bit number of Bit Allocation in Discrete controller control information source, channel
2, method is: determine the source encoding bit number according to the scope of channel bit error rate parameter value.Detailed process is:
As { the error rate<B
1K then
2=K
2N
B
1<=the error rate<B
2K then
2=K
2N-1
B
N-1<=the error rate is K then
2=K
21}
Wherein: B
1, B
2..., B
N-1, be respectively: 0.0005,0.001,0.0015,0.002,0.0025,0.003,0.0035.
Next utilizes system's information source, the total bit number of chnnel coding to deduct above-mentioned source encoding bit number, as the channel-encoded bit number.
If the comprehensive estimated value 0.0008 of the error rate of this moment, bit-error-rate estimator is estimating that the comprehensive error rate that obtains feeds back to transmitting terminal Bit Allocation in Discrete controller, and control also distributes information source, channel-encoded bit number, then is used for the bit number K of source encoding
2=202 bits are used for the bit number K of chnnel coding
4=286 bits.